Methods

Subjects

The questionnaires were collected at quarterly visits to the Eleanor and Lou Gehrig MDA/ALS Research Center at Columbia University. Responses from English-speaking patients with a probable or definite diagnosis of ALS using modified El Escorial criteria4 constitute the sample of convenience. Demographic and medical data were extracted from clinic charts. Data collection took place between January 2006 and December 2007. The Columbia University IRB approved the retrospective use of clinical data for research purposes.

Measures

These included the Fatigue Severity Scale (FSS) in which a score of >40 signifies clinically significant fatigue,5 Patient Health Questionnaire (PHQ)6 which generates an approximate diagnosis of clinical depression, the ALS Functional Rating Scale—Revised (ALSFRS-R)7 to assess severity of physical impairment, forced vital capacity (FVC)8 to measure respiratory capacity, and Progression Rate, which is calculated as the total possible ALSFRS-R score of 48 minus the actual score divided by months since symptom onset.9 10

Statistical analysis

Initially, the nine-item PHQ-9 was compared with a six-item version excluding three possibly ALS-related problems of psychomotor retardation, weight loss and sleep disturbance; the total score was prorated. The correlation between the two versions was sufficiently high (+0.94, p<0.001) that the full score was used in subsequent analyses.

Because of low frequencies, the 31 patients with either MDD (5+ symptoms) and 19 with minor depression (three or four symptoms) were combined in the “depressed” group. They did not differ on any sociodemographic or medical variable.

t Tests and ANOVAs were used to analyse relationships between the two dependent variables (fatigue and depression) and demographic and illness measures.

In comparing patient subgroups, MANOVA analyses, followed by pairwise comparisons when the overall F test was significant, and χ² tests for categorical data were performed. Stepwise regression and multinomial regression analyses were conducted to identify predictors of fatigue and depression, and a repeated-measures ANOVA was used to identify change and correlates over time.

In examination of patterns of change for patients seen on two occasions (n = 113) or three occasions (n = 65), fatigue and depression are reported without consideration of the overlap between them, as the samples become too small to partition in this manner.

The significance level for statistical tests was set at p = 0.05, and all tests were two-tailed.

Results

Sample

A total of 223 patients who met study inclusion criteria completed the fatigue and depression screens at least once, 113 on two occasions and 65 on three occasions, each about 3 months apart. The mean age was 61 (SD 12, range 32–88), 98 (56%) were male, and 181 (81%) were non-Hispanic white, with the remainder Asian, Black and Hispanic. The mean FVC was 73 (SD 23, range 7–129), and the mean ALSFRS-R was 33 (SD 8, range 5–46). Twenty-five per cent (43/173) of non-depressed patients and 34% (17/50) of depressed patients were prescribed antidepressants, but the purpose may have been to control sialorrhoea or insomnia rather than depression.

In order to evaluate characteristics of patients with clinically significant fatigue, depression (DSM-IV diagnosis of major or minor depressive disorder) or both, four subgroups were compared: fatigue only (n = 65, 29% of the total), depression only (n = 16, 7% of the total), both depression and fatigue (n = 34, 15% of the total) and neither (“comparison group,” n = 108, 48% of the total). Demographic and illness characteristics are shown in table 1.

Prevalence of fatigue at baseline

Of the 223 patients evaluated, 44% (99/223) had FSS scores >40, reflecting clinically significant fatigue; of these, 34% (34/99) were also depressed. As shown in table 1, patients with fatigue alone or fatigue accompanied by depression resembled each other but differed from the comparison group on indicators of ALS disease severity (FVC and ALSFRS-R) and rate of disease progression, reflecting greater impairment among patients with fatigue.

In addition, patients with fatigue with or without depression included a smaller proportion with bulbar onset (15% and 12% vs 50% of depressed patients and 25% of patients with neither condition, p = 0.009).

In a stepwise regression analysis with FSS score as the dependent variable, lower ALSFRS-R, non-bulbar onset, depression and more rapid ALS disease progression were all significant predictors of greater fatigue (F = 22.85, 194 df, p<0.001). Overall, greater fatigue was associated with greater ALS severity.

Persistence of fatigue over time

As shown in table 2, for the 39% (44/113) of patients with fatigue at time 1 who were assessed again, 75% (33/44) continued to report fatigue at time 2. In addition, of the 69 patients in this subset without fatigue at time 1, 36% (25/69) reported fatigue. Repeated-measures ANOVA revealed a significant increase in FSS over time (F = 9.15, df = 1/111, p = 0.003).

For patients assessed on three occasions, the patterns remained largely stable: 35% (23/65) had fatigue at time 1; of these, 87% (20/23) consistently reported fatigue at times 2 and 3, and 22% (14/65) of patients without fatigue at time 1 reported fatigue by time 3.

Prevalence of depression at baseline

At visit 1, 22% of patients (50/223) had symptoms consistent with MDD or minor depression (table 1). In the combined group of 50 depressed patients (with or without fatigue), there were no associations with demographic or illness severity measures.

In a multinomial regression analysis with depression diagnosis (major/minor/absent) as the dependent variable, and FSS score, ALSFRS-R, FVC, progression rate and bulbar/non-bulbar site of onset as independent variables, the only predictors of a depression diagnosis were higher fatigue scores (B = 0.073 (SE = 0.02), Wald = 11.621, 1 df, p = 0.001) and non-bulbar site of onset (B = −1.41 (SE = 0.57), Wald = 6.05, 1 df, p = 0.014). When repeated without entering fatigue as a predictor, none of the remaining variables was significant, indicating that the non-bulbar site was associated with fatigue rather than depression.

Persistence of depression over time

In the subset of 113 patients seen twice, 19% (21/113) were depressed at time 1. Of these, 48% (10/21) were also depressed at time 2, and an additional seven patients (6%) had new onset of depression at time 2, as can be seen in table 2. In the subset of 65 patients seen assessed three times, 17% (11/65) were depressed at time 1, of whom 36% (4/11) remained consistently depressed, while an additional 6% (4/65) became depressed at time 3. Overall, there was more variability in depression than fatigue across occasions.

What is the degree of overlap, and how do patients with both conditions differ from those with one or neither condition?

Table 1 compares demographic, medical and psychiatric measures for the four study subgroups at Time 1. There were no differences in age, gender or ethnicity between the four subgroups. Patients without either fatigue or depression had a greater respiratory capacity, less functional impairment and slower rate of progression than patients with fatigue, depression or both.

Discussion

Among patients screened for fatigue and depression in a multidisciplinary ALS clinic, fatigue was both frequent and persistent. The later stage of illness was associated with higher rates of fatigue. The finding of persistence of fatigue contrasts with fluctuating levels of fatigue reported in studies of patients with multiple sclerosis and psychiatric problems,11 12 although the time frame of our study was shorter. On the other hand, disease progression is much faster in ALS than most other diseases, and ALS is associated with more severe clinical weakness than either MS or psychiatric disorders, likely contributing to the persistence of fatigue in ALS.

Clinical depression was less common than fatigue and more likely to fluctuate over time. Sixty patients (27%) had been prescribed antidepressants at baseline, which may be a factor, although we have no measure of adherence, and the antidepressant may have been prescribed for symptoms other than depression. Moreover, there was no difference in rate of prescribed antidepressants among patients depressed at baseline (34%) and not depressed (25%) (χ² = 1.28, 1 df, p = 0.26).

While there is an overlap between symptoms of fatigue and depression (r = +.48, p<0.001), most patients with fatigue (65%) are not depressed, although most depressed patients have fatigue (68%). These data thus suggest that “pure” fatigue exists as an independent condition, as reported in other studies of both conditions (eg, Van der Linden et al12). These findings have implications for clinical management. If no treatable cause is identified, psychostimulant treatment may be warranted. Although efficacy data are sparse, our group found modafinil helpful in a placebo-controlled trial (response rates of 76% vs 14% in ITT analysis).13

Study limitations include the possibility of sample bias and reporting bias, since this was not a random sample. The period of observation was limited; further studies are needed to determine whether rates of fatigue continue to increase in ALS with further disease progression. Notwithstanding these limitations, these data suggest the need for routine enquiries about both fatigue and depression, as well as the importance of distinguishing between them.